optiroc / libsfx Goto Github PK

Following the "projects" link here gives a soft 404:

You may also check out the example projects to see how a project can be customized.

Hi (again... sorry),

I got several build issues on the example you provided in libSFX.

1/ Hello-Mode7: make output : http://sprunge.us/RaQh
2/ Mode7-Transform : http://sprunge.us/fThY
3/ X-GSU : http://sprunge.us/PSVO

For 1/ and 2/, no sfc was produced.
For 3/ a SuperFX.sfc is generated, but is not functionnal (pink background, tested with higan, snes9x, nosns) ?

Have a good day,
Respectfully,
ekyo

When creating a program with a map file to use for the rom and a map file to use for the SMP, the compiler gets confused and tries to use them interchangably. This causes obvious errors when the SMP doesn't have the right mapping details.

With only one map file, the CPU doesn't seem to want to steal the SMP's, but the SMP will try and use the main CPU's. This makes it impossible to create a program with a larger than normal ROM and also SMP overlays.

Hello, thank you for this excellent library, it helped me kickstart my dive into snes programming!

I load a background with

    LZ4_decompress Map, EXRAM, y
    VRAM_memcpy VRAM_MAP_LOC, EXRAM, y

    LZ4_decompress Tiles, EXRAM, y
    VRAM_memcpy VRAM_TILES_LOC, EXRAM, y

    CGRAM_memcpy 0, Palette, sizeof_Palette

and after the user presses the start button, another background will be loaded with

    LZ4_decompress Scr2Map, EXRAM, y
    VRAM_memcpy VRAM_MAP_LOC, EXRAM, y

    LZ4_decompress Scr2Tile, EXRAM, y
    VRAM_memcpy VRAM_TILES_LOC, EXRAM, y

    CGRAM_memcpy 0, Scr2Pal, sizeof_Scr2Pal

Unfortunately, this other background is garbled. The palette seems to load OK, but the map seems to be broken.

I have constructed a minimal example here (copy it into a subfolder of libsfx, e.g. into the examples folder):
https://github.com/josefcs/twopage-libsfx/

Forgive me, if I am missing something obvious here, but to me this seems like a bug.
Further pointers of what is going wrong would be very apreciated.

Thank you!

In the SMP-Play example the data in the SMP zeropage doesn't seem to get included in the main binary, and never gets loaded into APURAM.

Is this intended behavior? What would the procedure be to include the zeropage in the SMP binary?

Here's a screenshot from BizHawk's Hex Editor, showing the first few bytes of APURAM. The string Zero! that's included in the source file is not loaded: Imgur Link

Sorry if i'm completely misinterpreting the example. Thanks for help in advance.

Recently, I've written a (somewhat CPU-heavy) effect that renders into a buffer and uses WRAM_memset to clear the buffer each time. Since almost all the normal vblank DMA bandwidth is used to get the buffer into VRAM, I end up clearing the buffer mid-frame instead, where WRAM_memset becomes problematic on a real console.

Rather than touching VRAM, a safer option might be to just have a table of all 256 possible fill values in ROM and use that as the transfer source instead. This way, it can be safely used during active display.

The opcodes for the SuperFX blt and bge are swapped. libSFX assembles them as bge = 06xx and blt as 07xx, but these values should be reversed.

I have a LZ4-compressed file that crashes the decoder in SFX_LZ4_decompress/DecodeBlock. The cpu goes haywire and the stack loops around. I have permission to send the file privately, looking up your email to do so.

The shoulder button defines are the wrong way around. Curious that it's not been noticed til now, maybe so few games use them.

Master:

JOY_L                   = $0010
JOY_R                   = $0020

nocash docs:

15          1st        Button B          (1=Low=Pressed)
  14          2nd        Button Y
  13          3rd        Select Button
  12          4th        Start Button
  11          5th        DPAD Up
  10          6th        DPAD Down
  9           7th        DPAD Left
  8           8th        DPAD Right
  7           9th        Button A
  6           10th       Button X
  5           11th       Button L
  4           12th       Button R
  3           13th       0 (High)
  2           14th       0 (High)
  1           15th       0 (High)
  0           16th       0 (High)

.include "libSFX.i"

;-------------------------------------------------------------------------------
.segment "CODE"

.proc test
	RW_push set:a16
	RW_pull
.endproc

test.s(7): Error: Constant expression expected
../../libSFX/include/CPU.i(234): Note: Macro was defined here
test.s(7): Error: Constant expression expected
../../libSFX/include/CPU.i(238): Note: Macro was defined here
test.s(7): Error: Constant expression expected
../../libSFX/include/CPU.i(241): Note: Macro was defined here
test.s(7): Error: Constant expression expected
../../libSFX/include/CPU.i(244): Note: Macro was defined here
test.s(7): Error: Constant expression expected
../../libSFX/include/CPU.i(270): Note: Macro was defined here
../../libSFX/include/CPU.i(69): Note: Macro was defined here
"make": *** [.build/test.o] Error 1

Is there something non-obvious I might be doing wrong here? It seems like using any of the register size macros inside of a .proc causes problems with how those macros actually keep track of register state. I'd like to be able to tidy up my code/labels a bit but unfortunately this makes it difficult to do that without breaking any libSFX macros that use the RW macros (which is pretty much all of them...)